The breakability issues in the game are very impressive. Can you talk us through the process of implementing this and the problems you face when creating interactive scenery?
The breakability process starts in the game design department. They define which assets need to be breakable or bendable. During the production of Crysis 2 we had a level that contained all non-level specific props like trash bins, benches, air vents and other generic objects that was used to review these assets regarding breakability.
We then went through the individual levels themselves and defined the Action Bubbles in Crysis 2 that would benefit from breakability on any level-specific assets like the Grand Central Terminal asset only existing in one level.
We used this information to create a list of all assets that needed to be breakable and prioritized these assets based on their usage in the level. Things like cover objects that were used everywhere got a higher priority than assets that were only used once in the game. After that the assets were outsourced to the Crytek Kiev studio that had a dedicated group of five artists to only work on breakability. They cut up the meshes and sent them back to Frankfurt where I reviewed them and implemented the technical setup. If everything worked fine the meshes got implemented into the game.
The main problems we ran into were performance problems and missing technical features in our breakability pipeline. We had to reduce the number of pieces on some breakable assets since performance can drop significantly if all pieces are broken at once – imagine a grenade exploding in front of several pillars that are all breakable.
Other issues were visual glitches in the level of detail meshes or gameplay-relevant problems. How does the AI recognize when an object is broken; how does cover height get influenced by breakability; how many shots does it take to break a piece; which weapons can break a piece?
We came up with solutions for all of these issues, which proved a challenging process, requiring lots of work from all the departments concerned.
What do you feel are the most important elements to working as an environment artist and what are the most challenging things about the job?
Environment artists need both great artistic skills and a deep technical knowledge about the engine they work with. Understanding proportions,architecture and color theory are just as important as the technical aspects like the modularity or performance of your scene.
Being able to look at a concept painting and capture the essential features of the scene whilst understanding how to build it using the least amount of modules and textures is the most important skill an environment artist needs to learn.
The most challenging part of the job is usually working with the game designers or other departments and implementing their design ideas into your model, especially when it is based on a real-world environment. It can become quite tricky to implement certain gameplay metrics like cover heights or jump distances into your model without sacrificing visual quality.
In what ways do you avoid tiling issues when texturing large open areas and what are the typical maps you combine during the texturing process?
When creating the texture maps for large tiling surfaces we try to keep localized detail to a minimum in order to ensure the texture will tile well in the game. After that, we have several shaders that we can use in order to break up the tiling.
On game meshes we can use a blend shader that allows the blending of two different textures each using their own diffuse, normal and specular maps. The blending is controlled using vertex alpha, a height map and several shader attributes.
We can also use a dirt shader, which allows you to overlay a tiling texture of your choice onto your model. In this case, the normal map stays the same
but you can influence the specularity in the areas using the dirt shader. This is controlled via vertex alpha, too.
In addition to that we break up the tiling by using vertex colors on our meshes.
On terrain which is built in CryEngine we can paint several different materials and blend them into each other in real-time.
Level design does a great job of breaking up the tiling by placing additional decals on your model or cleverly distributing props on top of your objects as well.
Regarding the texturing process itself; we usually bake normal and ambient occlusion maps and use these to generate the diffuse and specular maps of our objects. We use a combination of hand-painting using photo overlays and a lot of manipulation of the baked maps. If the mesh resolution is dense enough we sometimes bake ambient occlusion into the vertex colors too.
Do you ever get to design the environments or do you generally rely on concept art for direction?
Since Crysis 2 plays in New York nearly all of our environments and levels are based on real-world locations. For these we rely a lot on photo reference we took when location scouting and a few bigger concept art pieces that define the general mood of the levels.
For the most iconic and recognizable pieces in the levels, like the City Hall asset in the level "Seat of Power” or the Castle Clinton asset in "Second Chance”, we created a lot of concept art to make sure we nailed the look and feel of these landmarks.
Other areas that are more generic, like the car park area in the opening of "Sudden Impact”, did not rely on any concept art at all and allowed me to design the area from start to finish.
Which environments have impressed you in other games you have not been involved with and why?
I was really impressed with the opening level in God of War 3. The fact that the levels are stunning environments and animated characters at the same time really floored me, especially considering how flawlessly animation and physics work together. The first time I saw something like that was in Shadow of the Colossus, which I still consider one of the best games of all time.
Gears of War had some impressive environments as well. The amount of detail each mesh contains is just breathtaking and the quality of the assets themselves are great – a true inspiration for anyone working as an environment artist.
Tell us a little about the background behind Black Mirror 2 and The Guild 2, and how these titles compared with working on the recent Crysis 2?
Black Mirror 2 is a classic point and click adventure game for the PC. It's based on static background images with 3D characters moving on top of them. Environment art production was fairly straightforward since the background plate was a flat 2D image file that is always seen from the same angle.
This simplified many things since you could treat the entire scene like a matte painting or a movie set, not having to worry about technical aspects. Anything that was not visible from the camera perspective did not need to be modeled and you could paint on the final image in post-production, easily modifying the look and feel of the scene.
The Guild 2 is a medieval life simulation and strategy game that uses the Gamebryo Engine. Regarding art production it was more comparable to Crysis 2 than Black Mirror 2. Each building is a 3D mesh with three different development stages based on each other. All buildings shared two 2048 x 2048 pixel textures for the entire game, which meant we had to be really efficient in our texturing and reuse a lot.
Each of these titles presented their own challenges that were vastly different from Crysis 2. The main difference is that both The Guild 2 and Black Mirror 2 were PC-only games that targeted a very specific audience, whereas Crysis 2 is a multiplatform title with much higher production values that needs to appeal to the mass market.